Operations needs a reliability strategy for multi-agent production: detect unhealthy agents, workflow failures, quality regressions, and behavioral drift, then trigger remediation. What should you design?
Select an answer to reveal the explanation.
Short Explanation
C is straight from the monitoring skills: agent health, workflow failures, trace correlation, drift detection, quality regression, and remediation. Multi-agent reliability is more than disk space. A ignores agent signals. B no logs means no diagnosis. D waiting for churn is not operations. Build metrics and traces that see coordination failures and quality drops, then automate safe remediation where patterns are known.
Full Explanation
Correct Answer — C
Reliability monitoring for multi-agent solutions covers agent health, workflow failures, trace correlation, behavioral drift, quality regression, and automated remediation for known failure patterns.
Why A is wrong: Infrastructure disk metrics alone miss agent cognitive and workflow failures.
Why B is wrong: Absence of logs prevents detection and forensics.
Why D is wrong: Customer churn is a lagging, unacceptable primary signal.
Exam tip: Drift + regression + remediation is the reliability triad on AI-500.